JOURNAL OF VIROLOGY, Jan. 1997, p. 650–656 Vol. 71, No. 1 0022-538X/97/$04.00ϩ0 Copyright ᭧ 1997, American Society for Microbiology

Association of the Parainfluenza Virus Fusion and Hemagglutinin-Neuraminidase Glycoproteins on Cell Surfaces

QIZHI YAO, XIAOLEI HU,† AND RICHARD W. COMPANS* Department of Microbiology and Immunology, Emory University School of Medicine, Atlanta, Georgia 30322

Received 5 August 1996/Accepted 2 October 1996

We previously observed that cell fusion caused by human parainfluenza virus type 2 or type 3 requires the expression of both the fusion (F) and hemagglutinin-neuraminidase (HN) glycoproteins from the same virus type, indicating that a type-specific interaction between F and HN is needed for the induction of cell fusion. In the present study we have further investigated the fusion properties of F and HN proteins of parainfluenza virus type 1 (PI1), type 2 (PI2), and type 3 (PI3), Sendai virus (SN), and simian virus 5 (SV5) by expression of their glycoprotein genes in HeLa T4 cells using the vaccinia virus-T7 transient expression system. Consistent with previous results, cell fusion was observed in cells transfected with homotypic F/HN proteins; with one exception, coexpression of any combination of F and HN proteins from different viruses did not result in cell fusion. The only exception was found with the closely related PI1 HN and SN HN glycoproteins, either of which could interact with SN F to induce cell fusion upon coexpression as previously reported. By specific labeling and coprecipitation of proteins expressed on the cell surface, we observed that anti-PI2 HN antiserum coprecipitated PI2 F when the homotypic PI2 F and PI2 HN were coexpressed, but not the F proteins of other paramyxoviruses when heterotypic F genes were coexpressed with PI2 HN, suggesting that the homotypic F and HN proteins are physically associated with each other on cell surfaces. Furthermore, we observed that PI3 F was found to cocap with PI3 HN but not with PI2 HN, also indicating a specific association between the homotypic proteins. These results indicate that the homotypic F and HN glycoproteins are physically associ- ated with each other on the cell surface and suggest that such association is crucial to cell fusion induced by paramyxoviruses.

Paramyxoviruses contain two surface glycoproteins: the fu- Recently, evidence localizing a region in HN which is im- sion protein (F) and the hemagglutinin-neuraminidase (HN), portant for promoting cell fusion has been obtained by analysis which form spikes on the viral envelope and are expressed on of chimeric proteins (9, 42, 46). By constructing several differ- the plasma membranes of infected cells. These two proteins ent chimeras from the HN genes of PI3 and Newcastle disease are directly involved in virus entry and cytopathology. The HN virus (NDV), the region which contains the transmembrane protein provides an attachment function, allowing virus to bind domain and the first 82 residues of the ectodomain in PI3 HN to sialic acid-containing receptors on the cell surface, and also was found to be important for its ability to interact with PI3 F exhibits a fusion-promoting activity (21), whereas the F protein proteins; the specificity of NDV HN for its homologous F is is known to mediate virus-cell fusion or cell-cell fusion (7, 8). determined by a similar domain (9). In studies of PI2 and The F protein is synthesized as an inactive precursor, F , which 0 simian virus 41 (SV41) HN proteins, a similar region near the is subsequently cleaved into the F and F subunits by a cellular 1 2 membrane-spanning domain and another region in the globu- trypsin-like protease which activates its fusion activity (14, 39). Cleavage of F results in exposure of a hydrophobic amino lar head of PI2 HN were reported to be involved in the cell fusion-promoting activity (46). A chimeric PI3-Sendai virus terminus on the F1 subunit (16, 19), which is thought to be directly involved in fusion activity (20). HN (SN HN) which contains only the 82-amino-acid region A number of previous studies have indicated that both gly- just outside the membrane-spanning segment of SN HN was coproteins, F and HN, participate in the fusion process (10, 15, sufficient for almost full fusion activity when coexpressed with 17, 28–30, 43, 45, 49). We previously demonstrated that cell the SN F protein (42). It is likely that these HN sequences fusion caused by parainfluenza virus type 2 (PI2) or type 3 could interact with sequences in the F protein which are lo- (PI3) glycoprotein requires expression of the F and HN pro- cated just external to the transmembrane region. Evidence has teins from the same virus type (17). No cell fusion was detected been obtained that a leucine zipper-like sequence near the upon coexpression of F and HN proteins originating from transmembrane domain (heptad repeat B region) of the F different virus types. These results support the conclusion that protein is necessary for syncytium formation in measles virus the HN protein plays a specific role in cell fusion, in addition and NDV (3, 37, 41). Furthermore, synthetic peptides corre- to its attachment function. A number of other recent reports sponding to the heptad repeat region B of paramyxovirus F also showed similar type-specific interactions among paramyxo- proteins have the potential to inhibit cell fusion (22, 33, 51). virus glycoproteins (2, 6, 13, 15, 25). To further understand the mechanism by which the paramyxovirus HN glycoprotein is involved in fusion activity, we have further investigated the interaction between F and * Corresponding author. Phone: (404) 727-5947. Fax: (404) 727- HN. By using the vaccinia virus-T7 transient expression system, 8250. we investigated the requirements for cell fusion mediated by † Present address: Laboratory of Viral Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Be- several paramyxoviruses, examined possible mechanisms which thesda, MD 20892. might be related to type-specific interactions between F and

650 VOL. 71, 1997 F AND HN FUSION COMPLEX OF PI / ON CELL SURFACES 651

HN, and investigated the physical association between the F TABLE 1. Cell fusion after coexpression of paramyxovirus F and and HN proteins on cell surfaces. HN proteinsa Cell fusionb after coexpression with: MATERIALS AND METHODS Protein PI1 HN PI2 HN PI3 HN SN HN SV5 HN Cells and viruses. HeLa T4 and HeLa cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) with 10% newborn calf serum (Hyclone Laboratories, PI2 F ϪϩϪϪϪ Inc., Logan, Utah). The vaccinia virus-T7 recombinant virus (WR strain) was obtained from Bernard Moss (National Institutes of Health) (11). PI3 F ϪϪϩϪϪ Recombinant plasmids. PI2 F and HN cDNA genes were cloned in pGEM-3 SN F ϩϪϪϩϪ and pGEM-3Zf(Ϫ) plasmids (Promega Biotech, Madison, Wis.), respectively, at SV5 F ϪϪϪϪϩ the SmaI site as described previously (17). The PI3 F and HN cDNA clones were a kindly provided by Mark Galinski. These two genes were subcloned into pGEM HeLa T4 cells were cotransfected with 5 ␮g of plasmid DNA following plasmid vectors. The PI1 HN cDNA gene was obtained from Yumiko Matsuoka vaccinia virus-T7 infection at 37ЊC for 1 h. Cell fusion was examined 8 to 24 h posttransfection. and was subcloned in pGEM-4. The pGEM-4-SNF and pGEM-4-SNHN cDNA b clones were kindly provided by Laurent Roux. The SV5 F and HN cDNA clones In positive experiments (ϩ), 60 to 70% of cells were found to be involved in were kindly provided by Robert A. Lamb and were subcloned in pGEM plas- syncytium formation, whereas no cell fusion was observed in the negative exper- mids. iments (Ϫ). Trypsin (4 ␮g/ml) was added to the medium of SN F-transfected Rabbit antisera. Rabbit antisera against PI2 or PI3 viruses and PI3 HN were cells. produced as described previously (17). Rabbit antisera against PI2 HN were generated as described by Ray and Compans (35). Monoclonal antibodies spe- cific for PI3 F and PI3 HN have been described previously (34, 36). Rabbit antiserum against Sendai virus was kindly provided by Laurent Roux. Rabbit incubated at 37 or 0ЊC, respectively, for 30 to 60 min. Then, primary antibody to antiserum against SV5 virus was kindly provided by Robert A. Lamb. the second antigen PI3 F and secondary antibody conjugated with fluorescein Transfection. Transfection of cells was performed as previously described (17). isothiocyanate (FITC) were added with each incubation interval for 30 min on Briefly, confluent monolayers of cells grown in 35-mm-diameter dishes were ice. Cells were washed with cold PBS-D and pelleted at 200 ϫ g after each infected with the vaccinia virus-T7 recombinant at a multiplicity of infection of 10 incubation. Cells were finally fixed in 1% paraformaldehyde and resuspended in and incubated at 37ЊC for 1 h. Cells were carefully washed with phosphate- 200 ␮l of PBS-D, mounted, and examined with a Zeiss LSM 410 invert confocal buffered saline (PBS) three times and then transfected with 5 ␮g of plasmid microscope. The specificity of antibodies to PI2 or PI3 HN was confirmed by DNA and 10 ␮l of lipofectin (Gibco BRL Life Technologies) in 2 ml of DMEM. immunofluorescence using cells expressing F proteins or heterotypic HN pro- Cells were incubated at 37ЊC for 16 h. Immunoprecipitation assays were then teins, which were found to be negative for fluorescence. performed. Radiolabeling and immunoprecipitation. Sixteen hours posttransfection, cells were starved in DMEM lacking methionine and cysteine for 30 min and labeled with 100 ␮Ci of [35S]methionine per ml for1hat37ЊC. Immunoprecipitation was RESULTS performed as described previously (50) and the radiolabeled polypeptides were analyzed by 10% reducing sodium dodecyl sulfate-polyacrylamide gel electro- Specificity of the interaction between F and HN. We previ- phoresis (SDS-PAGE) and subsequent autoradiography. ously demonstrated that using the vaccinia virus-T7 expression Biotinylation assay. Biotinylation of cell surface proteins was performed as system, cell fusion caused by PI2 or PI3 was only observed described by Lisanti et al. (23). Briefly, transfected cells were labeled with 100 ␮Ci of [35S]methionine per ml at 16 h posttransfection, washed with cold upon expression of F and HN proteins from the same virus 2ϩ 2ϩ PBS-CM [0.1 mM CaCl2, 1 mM MgCl2 in PBS-D (PBS lacking Mg and Ca )] type, which indicated that a type-specific interaction between three times and agitated with PBS-CM on ice for 30 min. After addition of the glycoproteins was involved in induction of cell fusion (17). NHS-SS-Biotin (Pierce, Rockford, Ill.) at a concentration of 0.5 mg/ml, cells To investigate the extent to which this type specificity extends were incubated on ice with agitation for 30 min. Cells were then lysed with mild lysis buffer (10 mM lauryl maltoside, 100 mM NaCl, 50 mM HEPES, pH 7.3) and to other paramyxoviruses, we have expressed the glycoproteins antibody was added. After addition of 10 ␮l of 50% protein A agarose beads and of three additional viruses, PI1, Sendai virus, and SV5 in HeLa incubation at 4ЊC for 2 h, cell lysates were washed with the lysis buffer and 10 ␮l T4 cells. By immunoprecipitation and/or indirect surface im- of 10% SDS was added, followed by boiling for 5 min. Supernatants were munofluorescence staining, the proteins encoded by all nine separated from agarose beads by addition of lysis buffer and centrifugation in a microcentrifuge. Ten microliters of 50% streptavidin agarose beads was added to clones presented in Table 1 were found to be expressed at high the supernatant, followed by incubation at 4ЊC for 2 h. Beads were then washed levels and transported to the surfaces of HeLa T4 cells (data and boiled, and biotinylated proteins were analyzed by 10% reducing SDS- not shown). In all F/HN coexpression combinations tested, PAGE and subsequent autoradiography. After scanning in the image, each band expression of F and HN from the same virus resulted in cell of interest was quantitated by using the IPLab gel program (Densitometry and Gel Analysis on the Macintosh; Signal Analytics Corporation). fusion, whereas pairs of F and HN proteins originating from Cell surface immunofluorescence. Cell surface double immunofluorescence different viruses were not observed to induce cell fusion, with analysis was performed as described by Paterson et al. (32) with some modifi- one exception. The single exception was observed in cells co- cations. In general, HeLa T4 cells were grown on glass coverslips in 24-well plates expressing SN F and PI1 HN, which is very closely related to and transfected as described above. Sixteen hours posttransfection, cells were washed with ice-cold PBS three times before being fixed with 1% paraformal- SN HN (27). These results with PI1 HN are consistent with dehyde in PBS. Rabbit anti-PI2HN (in cells cotransfected with heterotypic gly- observations reported by Bousse and coworkers (2). These coprotein genes) or rabbit anti-PI3 HN (in cells cotransfected with homotypic results demonstrate that the coexpression of HN and F pro- glycoprotein genes) antibody was added onto cell monolayers, and cells were teins from the same virus, or from very closely related viruses, then incubated at 4ЊC for 30 min. Cells were washed with ice-cold PBS three times, and then goat anti-rabbit rhodamine-conjugated antibody (Southern Bio- is required for cell fusion to occur, not only for parainfluenza technology Associates, Inc., Birmingham, Ala.) was added and cells were incu- virus type 2 and type 3 but also for the other paramyxoviruses bated at 4ЊC for 30 min. The cells were then incubated with mouse anti-PI3 F examined. antibody at 4ЊC for 30 min. After PBS washing, goat anti-mouse fluorescein- Sialic acid receptors are essential for cell fusion. We previ- conjugated antibody was added and cells were incubated at 4ЊC for another 30 min. Cell were then examined with a Zeiss LSM 410 invert confocal microscope. ously observed that addition of the lectin wheat germ aggluti- Cocapping assay. HeLa cells were grown in 100-mm-diameter cell culture nin to PI2 F-expressing cells did not result in cell fusion, indi- dishes in DMEM with 10% newborn calf serum. Antibody-induced capping was cating that the function of HN cannot be replaced by a lectin performed as described by Joseph and Oldstone (18) with modification. Briefly, which promotes cell attachment (17). We therefore investi- confluent monolayers of cells were cotransfected with PI3 F and PI3 HN or PI3 F and PI2HN. Sixteen hours posttransfection, cells were washed with warm gated whether the attachment activity of the HN protein to PBS-D three times and suspended with 2 mM EDTA at room temperature. After sialic acid is required for cell fusion by examining the effect of cells were washed with ice-cold PBS-D and pelleted by centrifugation at 1,000 treatment of transfected cells with purified Vibrio cholera neur- rpm, primary antibody to the PI3 HN or PI2HN was added followed by incuba- aminidase with four viruses: PI2, PI3, SNV, and SV5 as shown tion on ice for 30 min. Cells were then washed with ice-cold PBS-D and divided into two aliquots. After the secondary antibody conjugated with rhodamine in Fig. 1. In the cells not treated with neuraminidase, transfec- (rhodamine isothiocyanate [RITC]) was added, the two aliquots of cells were tion of F and HN resulted in extensive syncytium formation 652 YAO ET AL. J. VIROL.

FIG. 1. Effect of neuraminidase treatment of F/HN-cotransfected HeLa T4 cells on cell fusion. HeLa T4 cells were cotransfected with PI2 F and PI2 HN, PI3 F and PI3 HN, SN F and SN HN, and SV5 F and SV5 HN, as indicated. Cells were incubated at 37ЊC in the absence (top row) or presence (bottom row) of V. cholera neuraminidase (Gibco BRL Life Technology; protease content less than 0.1 mU/ml) at a concentration of 2.5 U/ml. Cell fusion was examined from 8 to 24 h posttransfection. Note that 4 ␮g of trypsin per ml was added to SN F-transfected cell medium.

(top photographs). In contrast, cell fusion was found to be V. cholera neuraminidase as shown in Fig. 2, lane 6. After almost completely inhibited when F/HN-transfected cells were neuraminidase treatment, PI2 F migrated to a position similar treated with neuraminidase (bottom photographs). The results to that of the PI2 F protein which was coexpressed with HN. indicate that the attachment of HN to sialic acid receptors is These results indicate that the sialic acid of the F protein can critical to cell fusion, since removal of these receptors results in be removed by coexpression with any of the HN proteins elimination of fusion activity. Moscona and Peluso (29, 30) tested, demonstrating that this phenomenon is unlikely to be have also reported evidence that interaction of the PI3 HN involved in the type-specific F/HN interactions required for protein with sialic acid receptors on neighboring cells is one of cell fusion. Further, the removal of sialic acid by HN demon- the essential steps in the fusion process. It is likely that this strates that the F protein is at least transiently present in close interaction facilitates effective contact of the F protein with the target cell membrane. An electrophoretic mobility change in F coexpressed with HN does not require a type-specific interaction. Our previous results showed that when PI2 F was expressed together with the HN protein, it always migrated faster than F expressed by itself, suggesting that sialic acid on the F protein was removed by the neuraminidase activity of HN (17). Since the neuramin- idase activities of various paramyxoviruses could differ in their specificities, it was of interest to determine whether this phe- nomenon is related to the type-specific interaction between F and HN that is involved in cell fusion. HeLa T4 cells were transfected with the PI2 F gene together with the HN gene of PI2, SN, PI1, or SV5 as shown in Fig. 2. As observed previ- ously, the PI2 F protein coexpressed with the PI2 HN protein (Fig. 2, lane 2) migrated faster than F expressed in the absence FIG. 2. Effect of neuraminidase activity of the HN protein on the electro- of HN (Fig. 2, lane 1). However, the PI2 F protein also mi- phoretic mobility of coexpressed F. HeLa T4 cells were transfected with PI2 F grated to the same position when coexpressed with other HN alone (lane 1), PI2 F with PI2 HN (lane 2), PI2 F with SN HN (lane 3), PI2 F with proteins, i.e., SN HN (lane 3), PI1 HN (lane 4), or SV5 HN PI1 HN (lane 4), PI2 F with SV5 HN (lane 5), and PI2 F plus neuraminidase (lane 6). Sixteen hours posttransfection, cells were labeled with [35S]methionine (lane 5). To confirm that such a faster migration is caused by at 37ЊC for 1 h, lysed with radioimmunoprecipitation assay buffer, and immuno- removal of sialic acid, PI2 F-transfected cells were treated with precipitated with rabbit anti-PI2 virus antiserum. VOL. 71, 1997 F AND HN FUSION COMPLEX OF PI / ON CELL SURFACES 653

surface proteins were immune precipitated with specific anti- bodies, solubilized, and then precipitated with streptavidin

agarose beads. The F1 protein of PI2 was found to be copre- cipitated by anti-PI2 HN antiserum when coexpressed with PI2

HN (lane 2). In contrast, although F1 proteins of SN F (lanes 3) or SV5 F (lanes 5) were shown to be expressed on cell surfaces when precipitated with antisera to the respective virus,

there was no coprecipitation of SN F1 (lane 4) or SV5 F1 (lane 6) when they were coexpressed with PI2 HN and precipitated by PI2 HN. In other experiments we observed that PI3 HN antibody was also able to coprecipitate the PI3 F protein but not PI2 F (not shown). To further determine whether the coprecipitation of HN and F requires interaction of the two proteins, cell lysates expressing PI2 F alone and lysates expressing PI2 HN alone were mixed and immunoprecipitated with PI2 HN antiserum.

No coprecipitation of PI2 F1 was observed with PI2 HN anti- serum (not shown). To determine what fraction of the surface F protein was coprecipitated with the anti-HN antibody, the amount of PI2 F protein detected on cell surfaces by antibody against PI2 virus and the amount coprecipitated by the PI2 HN antibody were quantitated and compared. The results indi- cated that about 80% of the PI2 F protein expressed on the cell surface was coprecipitated with PI2 HN antiserum. In order to exclude the possibility that failure to precipitate a heterotypic F is due to low levels of surface expression in the heterotypic F and HN combination, a cell surface immunofluorescence assay was performed. The levels of PI3 F on cell surfaces in cells coexpressing homotypic PI3 HN or heterotypic PI2 HN were found to be similar (data not shown). FIG. 3. Coprecipitation of F and HN proteins is type specific. (A) Specificity These results indicate that the PI2 and PI3 F and HN pro- of antiserum to the PI2 HN protein. HeLa T4 cells were transfected with PI2 F teins are physically associated with each other on cell surfaces, (lanes 2 and 3) or PI2 HN (lane 4) or infected with vaccinia virus-T7 (lane 1). Cells were labeled with [35S]methionine for1hat37ЊC at 16 h posttransfection. whereas heterotypic combinations of F and HN do not show a Cells were then lysed and immunoprecipitated with rabbit anti-PI2 virus anti- similar association. This physical association is likely to reflect serum (lane 2) or rabbit anti-PI2 HN antiserum (lanes 1 and 3 to 4). PI2 HN their functional interaction in the induction of cell fusion. protein was precipitated by the anti-PI2 HN antiserum (lane 4), whereas no PI2 Cocapping of homotypic F and HN glycoproteins. Many F protein was precipitated by this antibody (lane 3). (B) Biotinylation and coprecipitation of the F and HN proteins on the cell surface. HeLa T4 cells were viral glycoproteins have the potential to undergo antibody- cotransfected with PI2 F and PI2 HN (lanes 1 and 2), SN F and SN HN (lane 3), induced lateral redistribution on plasma membranes to form a SN F and PI2 HN (lane 4), SV5 F and SV5 HN (lane 5), or SV5 F and PI2 HN discrete cap on the cell surface (24). To further study the (lane 6) or transfected with PI2 F (lane 7), SN F (lane 8), or SV5 F (lane 9). At interactions of the F and HN proteins on the cell surface, 16 h posttransfection cells were labeled with [35S]methionine for3hat37ЊC, surface proteins were biotinylated with NHS-SS-Biotin, and then the biotinylated capping and cocapping assays were performed. If under con- proteins were immunoprecipitated with rabbit anti-PI2 virus antiserum (lane 1), ditions facilitating capping by specific antibody to HN the F rabbit anti-PI2 HN antiserum (lanes 2, 4, 6 to 9), rabbit anti-Sendai virus anti- proteins also migrate to the same position (cocapping) on the serum (lane 3), or rabbit anti-SV5 antiserum (lane 5). Trypsin (4 ␮g/ml) was cell surface, this would indicate that these two proteins are added to SN F-transfected cell medium during the labeling. Biotinylated proteins were finally precipitated with streptavidin agarose beads and analyzed by SDS- physically associated with each other. In contrast, if HN is PAGE. Double bands in the middle of the gel in all lanes are the background capped by specific anti-HN antibody while F still remains proteins from vaccinia virus infections. evenly distributed around the cell surface (no cocapping), this would indicate that the two proteins are not associated. Co- capping was examined by double staining with RITC labeling physical proximity to homotypic as well as heterotypic HN for the HN proteins and FITC labeling for the F proteins, proteins during intracellular transport or after arrival on the respectively. After cells were cotransfected with PI3 F and PI3 cell surface. HN for 16 h, cells were incubated with rabbit anti-PI3 HN Coprecipitation of the F protein by anti-HN antibody. The antiserum (at 0ЊC) for 30 min and RITC-conjugated goat anti- type-specific interaction of F and HN in induction of cell fusion rabbit immunoglobulin G (IgG) (at 37ЊC) for 45 min, respec- suggests that these two glycoproteins could be physically asso- tively, followed by incubation for 30 min with monoclonal ciated with each other, forming a functional complex which is antibody to PI3 F (at 0ЊC) and FITC-conjugated goat anti- critical for syncytium formation. To investigate the interactions mouse IgG (at 0ЊC). Figure 4a shows the red cap formed by PI3 of the F and HN proteins, a surface biotinylation assay was HN, while Fig. 4b shows a green cap formed by PI3 F. On the employed to selectively identify the proteins expressed on the surface of the doubly stained cells (Fig. 4c), it can be seen that cell surface (Fig. 3). The results in Fig. 3A demonstrate the the two membrane proteins are co-redistributed to form a specificity of the anti-PI2 HN antibody in immune precipita- unipolar cap in the same position. In contrast, we observed a tion assays. In Fig. 3B, HeLa T4 cells were cotransfected with lack of association of heterotypic combinations of F and HN PI2 F and PI2 HN (lanes 1 and 2), SN F and SN HN (lane 3), proteins on the cell surface by the cocapping assay. As shown SN F and PI2 HN (lane 4), SV5 F and SV5 HN (lane 5), and in Fig. 4d, most of the RITC-stained PI2 HN formed a cap, SV5 F and PI2 HN (lane 6) or singly transfected with PI2 F while FITC-stained PI3 F remained uniformly distributed on (lane 7), SN F (lane 8), or SV5 F (lane 9). The biotinylated the cell surface (Fig. 4e). Figure 4f shows the lack of colocal- 654 YAO ET AL. J. VIROL.

FIG. 4. Cocapping of F and HN proteins. By using the transient vaccinia virus-T7 expression system, HeLa cells were cotransfected with homotypic PI3 F and PI3 HN (a through c) or heterotypic PI3 F and PI2 HN (d through f). At 16 h posttransfection, cotransfected PI3 F and PI3 HN HeLa cells were first incubated with rabbit anti-PI3 HN antiserum (a through c); PI3 F and PI2 HN expressing HeLa cells were incubated with rabbit anti-PI2 HN antibody (d through f) at 0ЊC for 30 min and then with RITC-conjugated goat anti-rabbit IgG at 37ЊC for 45 min. Cells were then incubated with monoclonal antibody to PI3 F at 0ЊC for 30 min and subsequently with FITC-conjugated goat anti-mouse IgG at 0ЊC for 30 min. Cells were examined with a Zeiss LSM confocal microscope. PI3 F and PI3 HN cotransfected cells showed rhodamine staining of a red cap formed by PI3 HN (a), fluorescein staining showing cocapping of PI3 F (b), and an image corresponding to PI3 F and PI3 HN in the same focal plane is shown in panel c. The yellow cap shows that both the PI3 F and PI3 HN proteins cocap in the same position. PI3 F and PI2 HN cotransfected cells showed rhodamine staining of a red cap formed by PI2 HN (d), fluorescein staining of PI3 F uniformly distributed on the cell surface (e), and an image corresponding to PI3 F and PI2 HN in the same focal plane is shown in panel f. There was no cocapping observed with heterotypic PI3 F and PI2 HN on the cell surface. ization of the two images. These results indicate that F and HN induce cell fusion is an exception, but it can be explained by the proteins of the same virus type laterally move together on the fact that the HN proteins of PI1 and SN viruses are the most cell surface under conditions of antibody-induced capping, closely related among paramyxoviruses with primary sequence whereas heterologous F and HN proteins fail to show a similar identity of 83.1% (27). The structure of HN of the two viruses association. would be predicted to be very similar, as shown by a nearly identical hydropathicity profile and by conservation of struc- DISCUSSION tural elements, including all 17 cysteines, 26 of 27 prolines, and 29 of 31 glycines (12). Recently, it was reported in several Our previous finding that the HN protein is not only re- studies that certain heterotypic glycoprotein combinations quired for cell fusion but that coexpressed F and HN proteins could also induce cell fusion (1, 2, 46). Our result with PI1 is in also need to be derived from the same virus type demonstrated agreement with the report by Bousse and coworkers (2). These that the F and HN glycoproteins are involved in a type-specific results are thought to reflect a high level of sequence similarity interaction which promotes cell fusion. Therefore, the mech- in the HN proteins or differences in requirements of the F anism of cell fusion seems to be more complex than had been proteins for induction of cell fusion. previously believed because of the nature of the specific in- In the present study we found that the F protein of PI2 was volvement of the HN protein in the fusion process. On the coprecipitated by antibody to PI2 HN, indicating that the two basis of our previous findings, several alternative mechanisms of paramyxovirus F-HN interaction were considered. These proteins are tightly associated with each other. Furthermore, include the role of the binding of HN to sialic acid receptors, the anti-PI2 HN antibody was only found to coprecipitate the the effect of HN neuraminidase on removal of sialic acid from homotypic F protein, indicating that the homotypic F and HN F, and the possible formation of an F/HN glycoprotein com- proteins are physically associated on the cell surface. The re- plex. The present results provide evidence for the formation of sults suggest that a type-specific fusion complex may form on such a complex on cell surfaces. the cell surfaces in which the two proteins physically contact We demonstrated by coexpression of homotypic and hetero- each other. In contrast, no such interaction was detected be- typic F/HN proteins that the type-specific F-HN interaction is tween the heterotypic F and HN combinations on the cell not only required for parainfluenza virus types 2 and 3 but also membrane. Similarly, there is no coprecipitation of homotypic occurs with other paramyxoviruses, including Sendai virus and F and HN by HN antibody when cell lysates expressing the SV5. These results are consistent with our previous conclusion individual proteins were mixed together. Malvoisin and Wild that an interaction between F and HN of the same virus sero- (25) reported that measles virus F and HA proteins expressed type is crucial to cell fusion induced by paramyxoviruses and in HeLa cells infected with vaccinia virus recombinants can be indicate that they share a common mechanism in the fusion coimmunoprecipitated in the presence of the cross-linking re- process. The finding that PI1 HN can interact with SN F to agent DSP. We found that the coprecipitation of the F and HN VOL. 71, 1997 F AND HN FUSION COMPLEX OF PI / ON CELL SURFACES 655 proteins of PI2 virus was detected even without the addition of contrast, paramyxovirus fusion proteins do not have receptor a cross-linking reagent. Additional evidence for homotypic F binding activity. However, our data indicate that cell fusion and HN association was obtained by a cocapping assay, show- induced by paramyxovirus also involves formation of a complex ing that combinations of F and HN proteins which interact to in which F and HN are physically associated with each other. It induce cell fusion laterally move together when redistributed will be of interest to determine whether the formation of such on the cell surface by bivalent antibody. Only if the two pro- a complex may result in a conformational change of the F teins are physically linked together would migration of one protein, similar to the conformational changes of fusion pro- protein be expected to cause the other protein to redistribute teins of other enveloped viruses such as influenza virus, vesic- to the same position. Consistent with the coprecipitation re- ular stomatitis virus, or Semliki Forest virus, which are trig- sults, the cocapping of the two glycoproteins was found to gered by a low-pH environment (26, 31). The resulting change occur only with F and HN of the same virus type; heterotypic may lead to further exposure of the amino terminus of the F1 combinations of F and HN were not found to laterally move subunit, as occurs in the influenza virus HA protein (4, 5). together. Alternatively, the functional interaction of the two proteins Recently, an alternative explanation of the type specificity of may depend on specific lateral interactions between F and HN, F-HN interaction was proposed by Tanaka and coworkers (44). leading to assembly into a patch-like structure on the plasma They reported that PI3 F-KDEL, a mutant F protein contain- membrane. Such a domain may lead to the effective contact of ing a signal for retention in the endoplasmic reticulum, can F with target membranes, which results in membrane fusion, down-regulate PI3, SV5, and SN HN and measles virus H whereas individual F subunits may be unable to mediate fusion expression, suggesting that the down-regulation of heterotypic efficiently because they are randomly distributed on the cell HN expression may result in the serotype-specific restriction of surface. syncytium formation. They proposed that a transient F-HN interaction could induce a conformational change in the F ACKNOWLEDGMENTS protein. We previously observed, however, that coexpression We thank Mark Galinski, Robert A. Lamb, and Laurent Roux for of a heterotypic PI3 F protein in cells coexpressing PI2 F and providing cDNA clones. We also thank Lawrence R. Melsen for pho- PI2 HN did not block the cell fusion induced by the homotypic tography and Tanya Cassingham for assistance in preparing the manu- proteins (17). Therefore, it seems unlikely that the heterotypic script. F proteins have a dominant negative effect on fusion activity. This study was supported by research grant AI 12680 from the Paramyxoviruses also encode a matrix protein which forms a National Institute of Allergy and Infectious Diseases, NIH. Q. Yao was dense layer which underlies the viral membrane. It has been supported by the Public Health Service, National Research Service reported that in Sendai virus-infected cells, the M protein Award AI 07470 from the National Institutes of Health. binds independently to either the F or the HN protein, which REFERENCES is considered to be an important step in virus assembly (38). 1. Bagai, S., and R. A. Lamb. 1995. Quantitative measurement of paramyxovi- An association with M could thus play an important role in HN rus fusion: differences in requirements of glycoproteins between simian virus and F association. However, the present results support the 5 and human parainfluenza virus 3 or Newcastle disease virus. J. Virol. conclusion that such an association can occur in the absence of 69:6712–6719. M, and many studies have obtained evidence that expression of 2. Bousse, T., T. Takimoto, W. L. Gorman, T. Takahashi, and A. Portner. 1994. Regions on the hemagglutinin-neuraminidase proteins of human parainflu- M is not required for paramyxovirus fusion activity. We have enza virus type 1 and Sendai virus important for membrane fusion. Virology also observed that PI2 F cytoplasmic domain truncation mu- 204:506–514. tants promote efficient cell fusion when coexpressed with PI2 3. Buckland, R., E. Malvoisin, P. Beauverger, and F. Wild. 1992. A leucine HN (50). It will be of interest to determine whether the asso- zipper structure present in the measles virus fusion protein is not required for its tetramerization but is essential for fusion. J. Gen. Virol. 73:1703–1707. ciation of the M protein differs in the case of homotypic or 4. Bullough, P. A., F. M. Hughson, J. J. Skehel, and D. C. Wiley. 1994. Structure heterotypic F or HN proteins and in truncated forms of the F of influenza haemagglutinin at the pH of membrane fusion. Nature (Lon- protein. don) 371:37–43. Recently, some progress has been made toward defining the 5. Carr, C. M., and P. S. Kim. 1993. A spring-loaded mechanism for the conformational change of influenza hemagglutinin. Cell 73:823–832. regions involved in the type-specific interaction of F and HN. 6. Cattaneo, R., and J. K. Rose. 1993. Cell fusion by the envelope glycoproteins Sergel and coworkers (40) reported that the HN fusion pro- of persistent measles viruses which caused lethal human brain disease. J. Vi- motion activity can be separated from its hemagglutinating rol. 67:1493–1502. activity. Several studies on determination of the regions of the 7. Choppin, P. W., and A. Scheid. 1980. The role of viral glycoproteins in adsorption, penetration and pathogenicity of viruses. Rev. Infect. Dis. 2:40– proposed F and HN interactions (9, 42, 46) have concluded 61. that a region in the extracellular domain adjacent to the trans- 8. Choppin, P. W., and R. W. Compans. 1975. Reproduction of paramyxovi- membrane domain of the HN protein is important for inter- ruses, p. 95–178. In R. R. Wagner and H. Fraenkel-Conrat (ed.), Compre- action with the respective F protein in promoting cell fusion. hensive virology. Plenum Press, New York. 9. Deng, R., Z. Wang, A. M. Mirza, and R. M. Iorio. 1995. Localization of a Using chimeric proteins, a 45-amino-acid cysteine-rich seg- domain on the paramyxovirus attachment protein required for the promo- ment has been identified in the measles virus F protein that can tion of cellular fusion by its homologous fusion protein spike. Virology transfer specificity for its attachment protein to the F protein 209:457–469. of canine distemper virus (48). These results all support the 10. Ebata, S. N., M.-J. Cote, C. Y. Kang, and K. Dimock. 1991. The fusion and hemagglutinin-neuraminidase glycoproteins of human parainfluenza virus 3 proposal that type-specific association between the homotypic are both required for fusion. Virology 183:437–441. F and HN plays an important role in the cell fusion process. 11. Fuerst, T. R., E. G. Niles, F. W. Studier, and B. Moss. 1986. Eukaryotic Taken together, the present study further demonstrates the transient-expression system based on recombinant vaccinia virus that syn- requirement of F and HN in paramyxovirus-induced cell fu- thesizes bacteriophage T7 RNA polymerase. Proc. Natl. Acad. Sci. USA 83:8122–8126. sion, the requirement that both glycoproteins have to be from 12. Gorman, W. L., D. S. Gill, R. A. Scroggs, and A. Portner. 1990. The hem- the same or a very closely related virus type, and the require- agglutinin-neuraminidase glycoproteins of human parainfluenza virus type 1 ment for attachment of HN to sialic acid in addition to its and Sendai virus have high structure-function similarity with limited anti- specific role in cell fusion. In most enveloped viruses, a single genic cross-reactivity. Virology 175:211–221. 13. Heminway, B. R., Y. Yu, and M. S. Galinsky. 1994. Paramyxovirus mediated fusion protein is sufficient for fusion activity; however, the cell fusion requires co-expression of both the fusion and hemagglutinin- same molecule is also responsible for receptor binding (47). In neuraminidase glycoproteins. Virus Res. 31:1–16. 656 YAO ET AL. J. VIROL.

14. Homma, M., and M. Ohuchi. 1973. Trypsin action on the growth of Sendai viruses. EMBO J. 14:5524–5531. virus in tissue culture cells. III. Structural differences of Sendai viruses grown 34. Ray, R., V. E. Brown, and R. W. Compans. 1985. Glycoproteins of human in eggs and tissue culture cells. J. Virol. 12:1457–1463. parainfluenza type 3 virus: characterization and evaluation as a subunit 15. Horvath, C. M., R. G. Paterson, M. A. Shaughnessy, R. Wood, and R. A. vaccine. J. Infect. Dis. 152:1219–1230. Lamb. 1992. Biological activity of paramyxovirus fusion proteins: factors 35. Ray, R., and R. W. Compans. 1987. Glycoproteins of human parainfluenza influencing formation of syncytia. J. Virol. 66:4564–4569. virus type 3: Affinity purification, antigenic characterization, and reconstitu- 16. Hsu, M.-C., A. Scheid, and P. W. Choppin. 1981. Activation of the Sendai tion into lipid vesicles. J. Gen. Virol. 68:409–418. virus fusion protein (F) involved a conformational change with exposure of 36. Ray, R., and R. W. Compans. 1986. Monoclonal antibodies reveal extensive a new hydrophobic region. J. Biol. Chem. 256:3557–3563. antigenic differences between the hemagglutinin-neuraminidase glycopro- 17. Hu, X., R. Ray, and R. W. Compans. 1992. Functional interactions between teins of human and bovine parainfluenza 3 viruses. Virology 148:232–236. the fusion protein and hemagglutinin-neuraminidase of human parainflu- 37. Reitter, J. N., T. Sergel, and T. G. Morrison. 1995. Mutational analysis of the enza viruses. J. Virol. 66:1528–1535. leucine zipper motif in the Newcastle disease virus fusion protein. J. Virol. 18. Joseph, B. S., and M. B. A. Oldstone. 1974. Antibody-induced redistribution 69:5995–6004. of measles virus antigens on the cell surface. J. Immunol. 113:1205–1209. 38. Sanderson, C. M., H.-H. Wu, and D. P. Nayak. 1994. Sendai virus M protein 19. Kohama, T., W. Garten, and H.-D. Klenk. 1981. Changes in conformation binds independently to either the F or the HN glycoprotein in vivo. J. Virol. and charge paralleling proteolytic activation of Newcastle disease virus gly- 68:69–76. coproteins. Virology 111:364–376. 39. Scheid, A., and P. W. Choppin. 1974. Identification and biological activities 20. Lamb, R. A. 1993. Paramyxovirus fusion: a hypothesis for changes. Virology of paramyxovirus glycoproteins. Activation of cell fusion, hemolysis, and 197:1–11. infectivity by proteolytic cleavage of an inactive precursor protein of Sendai 21. Lamb, R. A., and D. Kolakofsky. 1996. Paramyxoviridae: the viruses and their virus. Virology 57:475–490. replication, p. 1177–1204. In B. N. Fields, D. M. Knipe, and P. M. Howley 40. Sergel, T., L. W. McGinnes, M. E. Peeples, and T. G. Morrison. 1993. The (ed.), Virology, vol. 1. Lippincott-Raven, New York. attachment function of the Newcastle disease virus hemagglutinin-neuramin- 22. Lambert, D. M., S. Barney, A. L. Lambert, K. Guthrie, R. Medinas, D. E. idase protein can be separated from fusion promotion by mutation. Virology Davis, T. Bucy, J. Erickson, G. Merutka, and S. R. J. Petteway. 1996. 193:717–726. Peptides from conserved regions of paramyxovirus fusion (F) proteins are 41. Sergel, T. G., C. McQuain, and T. G. Morrison. 1994. Mutations in the fusion potent inhibitors of viral fusion. Proc. Natl. Acad. Sci. USA 93:2186–2191. peptide and heptad repeat regions of the Newcastle disease virus fusion 23. Lisanti, M. P., A. Le Bivic, M. Sargiacomo, and E. Rodriguez-Boulan. 1989. protein block fusion. J. Virol. 68:7654–7658. Steady-state distribution and biogenesis of endogenous Madin-Darby canine 42. Tanabayashi, K., and R. W. Compans. 1996. Functional interaction of kidney glycoproteins: evidence for intracellular sorting and polarized cell paramyxovirus glycoproteins: identification of a domain in Sendai virus HN surface delivery. J. Cell Biol. 109:2117–2127. which promotes cell fusion. J. Virol. 70:6112–6118. 24. Lydy, S. L., S. Basak, and R. W. Compans. 1990. Host cell-dependent lateral 43. Tanabayashi, K., K. Takeuchi, K. Okazaki, M. Hishiyama, and A. Yamada. mobility of viral glycoproteins. Microbial. Pathol. 9:375–386. 1992. Expression of mumps virus glycoproteins in mammalian cells from 25. Malvoisin, E., and T. F. Wild. 1993. Measles virus glycoproteins: studies on cloned cDNAs: both F and HN proteins are required for cell fusion. Virology the structure and interaction of the hemagglutinin and fusion proteins. 187:801–804. J. Gen. Virol. 74:2365–2372. 44. Tanaka, Y., B. R. Heminway, and M. S. Galinski. 1996. Down-regulation of 26. Marsh, M., and A. Helenius. 1989. Viral entry into animal cells. Adv. Virus paramyxovirus hemagglutinin-neuraminidase glycoprotein surface expres- Res. 36:107–151. sion by a mutant fusion protein containing a retention signal for the endo- 27. Matsuoka, Y., R. Ray, and R. W. Compans. 1990. Sequence of the hemag- plasmic reticulum. J. Virol. 70:5005–5015. glutinin-neuraminidase gene of human parainfluenza virus type 1. Virus Res. 45. Taylor, J., S. Pincus, J. Tartaglia, C. Richardson, G. Alkhatib, D. Briedis, M. 16:107–113. Appel, E. Norton, and E. Paoletti. 1991. Vaccinia virus recombinants ex- 28. Morrison, T., C. McQuain, and L. McGinnes. 1991. Complementation be- pressing either the measles virus fusion or hemagglutinin glycoprotein pro- tween a virulent Newcastle disease virus and a fusion protein gene expressed tect dogs against canine distemper virus challenge. J. Virol. 65:4263–4274. from a retrovirus vector: requirement for membrane fusion. J. Virol. 65:813– 46. Tsurudome, M., M. Kawano, T. Yuasa, N. Tabata, M. Nishino, H. Komada, 822. and Y. Ito. 1995. Identification of regions on the hemagglutinin-neuramini- 29. Moscona, A., and R. Peluso. 1992. Fusion properties of cells infected with dase protein of human parainfluenza virus type 2 important for promoting human parainfluenza virus type 3: receptor requirements for viral spread and cell fusion. Virology 213:190–203. virus-mediated membrane fusion. J. Virol. 66:6280–6287. 47. White, J. M. 1992. Membrane fusion. Science 258:917–924. 30. Moscona, A., and R. W. Peluso. 1991. Fusion properties of cells persistently 48. Wild, T. F., J. Fayolle, P. Beauverger, and R. Buckland. 1994. Measles virus infected with human parainfluenza virus type 3: participation of hemagglu- fusion: role of the cysteine-rich region of the fusion glycoprotein. J. Virol. tinin-neuraminidase in membrane fusion. J. Virol. 65:2773–2777. 68:7546–7548. 31. Ohnishi, S.-I. 1988. Fusion of viral envelopes with cellular membranes. Curr. 49. Wild, T. F., E. Malvoisin, and R. Buckland. 1991. Measles virus: both the Top. Membr. Transp. 29:257–296. hemagglutinin and fusion glycoproteins are required for fusion. J. Gen. 32. Paterson, R. G., M. S. Shaughnessy, and R. A. Lamb. 1989. Analysis of the Virol. 72:439–442. relationship between cleavability of paramyxovirus fusion protein and length 50. Yao, Q., and R. W. Compans. 1995. Differences in the role of the cytoplasmic of the connecting peptide. J. Virol. 63:1293–1301. domain of human parainfluenza virus fusion proteins. J. Virol. 69:7045–7053. 33. Rapaport, D., M. Ovadia, and Y. Shai. 1995. A synthetic peptide correspond- 51. Yao, Q., and R. W. Compans. 1996. Peptides corresponding to the heptad ing to a conserved heptad repeat domain is a potent inhibitor of Sendai repeat sequences of human parainfluenza virus fusion protein are potent virus-cell fusion: an emerging similarity with functional domains of other inhibitors of virus infection. Virology 223:103–112.